INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

VP16 is encoded by the UL48 gene of herpes simplex virus type 1 (HSV-1) and is an essential structural protein, assembled into the tegument of the virion at approximately 1,200 to 1,500 molecules per particle (14). It is also a potent activator of the transcription of viral immediate-early (IE) genes (2, 3, 32). Transcriptional activation is initiated by the recruitment of VP16, together with two cellular proteins, Oct-1 (9, 30, 33, 38) and host cell factor (HCF) (17, 50, 51), into a multicomponent complex formed on regulatory sites (TAATGARAT motifs) present within each of the IE gene promoters (for reviews, see references 29 and 49). Previous analyses from several laboratories are consistent with a model of VP16 whereby the functions of recruitment to the DNA binding complex and transcriptional activation are separate activities located in two discrete domains (4, 10, 43). An amino-terminal domain, refined to within residues 49 to 390, is involved in the binding of VP16 to HCF and the recruitment of this binary complex to the Oct-1-DNA complex (11). Analysis by limited proteolysis demonstrated that the region at about residue 370 within this domain is present in a surface-exposed loop (13); results from site-directed mutagenesis showed that, while complex formation is sensitive to alterations in other regions, key residues involved in interactions with Oct-1 and HCF are located within residues 360 to 390 (1, 11, 21, 48).

VP16 extends for another 100 residues beyond the C-terminal boundary of the domain required for complex formation, and this C-terminal extension is highly enriched in acidic amino acids (6, 31). This region encompasses determinants required for transcriptional activation, since C-terminal truncations or insertions within the intact protein abolished activation without having any detectable effect on complex formation (1, 10, 35, 48). However, most studies of the determinants involved in transcriptional activation per se have been performed in the context of fusion proteins in which the C-terminal region has been fused to a heterologous DNA binding domain, e.g., that from the yeast GAL4 protein, and activation has been studied with target promoters containing GAL4 recognition sites (4, 37). From such studies, the C-terminal region has been generally recognized as a physical and functional domain which can be split broadly into two subdomains, the H1 or N region (residues 410 to 452) and the H2 or C region (residues 453 to 490) (35, 43, 46). Although the net negative charge in the C-terminal region contributes to activation, the pattern of hydrophobic and aromatic residues appears to be more critical, with particularly important residues being phenylalanines at position 442 in H1 (N) and at positions 473, 475, and 479 in H2 (C) (5, 35, 46). The number of targets proposed to bind to the VP16 activation domain is confusingly large and includes members of the basal transcription initiation complex (18, 23, 39), mediator proteins and RNA polymerase II holoenzyme (15, 19), histone acetylases (44), and many other members of the transcriptional apparatus, and it is presently difficult to reconcile a role for all of these factors in a physiologically relevant way (for a review, see reference 41). Moreover, comparison with the homologues of VP16 from other alphaherpesviruses has emphasized some of the difficulties in understanding the detailed mechanism of activation.

For example, within the VP16 homologues from bovine herpesvirus 1 (BHV-1), equine herpesvirus 1 (EHV-1), and varicella-zoster virus (VZV), the N terminus is well conserved, but there is significant divergence in the C terminus among these proteins. Indeed, based on the known requirements within VP16-H and the differences found within the C termini, it was predicted that, e.g., the EHV-1 homologue might not activate IE gene expression (40). However, each of the VP16 species, including that of EHV-1, has been reported to activate IE gene expression (8, 22, 25, 28, 34). The C terminus of the EHV-1 protein (termed VP16-E for ease of reference in this work) exhibits some homology with that of VP16-H but appears to be a truncated version and does not have the same preponderance of negatively charged residues. However, this region of VP16-E has been shown to be required for its transcriptional activity (8).

As part of a program to clarify the mechanisms involved in IE gene activation, we sought to directly compare VP16-H and VP16-E, particularly with respect to the involvement of C-terminal regions in complex formation and transactivation. Our results indicate some significant differences between the two proteins, since a determinant within the C terminus of VP16-E is required for the assembly of the complex with Oct-1 and HCF. Paradoxically, this region, while representing a selectively conserved segment within the otherwise diverged C termini, is dispensable in VP16-H. Moreover, we show that the activities of GAL4 fusion proteins do not reflect the activities of the intact parental proteins, indicating that the mode of recruitment of the activation domain may be crucial for activation. Key determinants required for activation in VP16-E, and possibly VP16-H, may involve interactions between regions of the C terminus and the N terminus rather than discrete domains with independent functions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells, transfections, and CAT assays. COS-1 and Vero cells were grown in Dulbecco's modified minimal essential medium containing 10% newborn calf serum. Transfections were performed by the calcium phosphate method with various amounts of expression plasmids made up to 2 µg with pUC19 DNA as described previously (11). Cells were harvested approximately 40 h after transfection and assayed for chloramphenicol acetyltransferase (CAT) activity exactly as described previously (11).

Plasmids. VP16-H cloned in pcDNA1 (Invitrogen) was transferred from the parental construct (26) into pcDNA1/amp as a HindIII-EcoRI fragment. VP16-E was initially produced by PCR amplification of gene 12 from EHV-1 strain Ab1 in vector GE126 (8). Alignment of the coding sequences of various VP16 species indicated that in this original construct, translation of VP16-E may have initiated from an upstream in-frame methionine, yielding a protein with an extra 30 residues. VP16-E was therefore recloned by PCR amplification with primers containing EcoRI and XbaI sites at the 5' and 3' ends, respectively. The 5' primer was positioned to amplify from the next available methionine, yielding a protein about 30 residues shorter than the original one. The fragment was inserted between the EcoRI and XbaI sites of pcDNA1/amp to yield the expression vector for wild-type VP16-E, pcDNA1/amp.VP16-E. (The lack of 30 residues from the original construct had no deleterious effect on activity but appeared to increase the efficiency of expression.) Mutations in VP16-E were introduced by subcloning from plasmids GE130, GE134, GE135, GE136, and GE137, which contain mutant VP16-E species (8), by swapping the C-terminal BglII-HpaI fragment of the parental vector with the corresponding fragment from each of the mutants in the GE series of plasmids to yield VP16-E.(1-425), VP16-E.(1-445), VP16-E.(444), VP16-E.(442-444), and VP16-E.(442-445).

Gel retardation assays. Binding reactions were carried out as previously described (16) using 0.02 µl (0.2 ng) of the purified POU domain, in vitro-translated HCF, and 1 ng of the end-labeled probe. VP16 species were supplied by in vitro translation (TnT; Promega) or from soluble extracts of transfected cells prepared as described previously. The probes encompassed the E1 octamer site of the EHV-1 IE gene promoter (AGCTGAGGAGACGCATGCAGATGAGATGTGCATCGAGG) (8) or the octamer motif at position 160 of the HSV-1 IE110 gene promoter (TAAT24) as previously described (CCATGGAGATCTCGTGCATGCTAATGATATTCTTCCATGG) (underlined sequences indicate the octamer site within the probe) (47). Poly(dI-dC) was routinely added to the mixture at 1 µg per reaction to reduce nonspecific binding. The mixture was incubated with the labeled probe for 15 min, and the complexes were resolved in nondenaturing 6% polyacrylamide (acrylamide-bisacrylamide, 37:5) gels in 0.5× Tris-borate-EDTA (TBE). Electrophoresis was performed at a constant voltage of 200 V for 90 min. The gels were dried, and complexes were detected by autoradiography.

Immunofluorescence. COS-1 cells to be processed for immunofluorescence were seeded at 1.25 × 10⁵ cells per well in six-well cluster plates (Costar) on 25-mm glass coverslips. Approximately 40 h after transfection, cells were washed with phosphate-buffered saline (PBS), fixed for 15 min with ice-cold methanol, and blocked in PBS containing 10% calf serum (blocking solution) for 20 min. Monoclonal antibody against the SV5 tag was added in the same solution (1:2,000) for 20 min. VP16-H was detected by using the monoclonal antibody LP1 as previously described (20). For secondary antibodies, fluorescein isothiocyanate-conjugated anti-mouse immunoglobulin G (Vector Laboratories) was used at a dilution of 1:100 and tetramethyl rhodamine isothiocyanate-conjugated anti-rabbit immunoglobulin G (Sigma) was used at a dilution of 1:200. These antibodies were added in blocking solution and incubated for 20 min. After three 5-min washes in PBS, the coverslips were mounted in Vector Shield (Vector Laboratories) and visualized using either a Bio-Rad MRC600 confocal microscope or a Zeiss LSM 410 confocal microscope. Images were processed with Adobe Photoshop software.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Requirements for a conserved determinant in the extreme C terminus. A schematic summary of the relationship between the VP16-H and VP16-E proteins is shown in Fig. 1a. A more detailed alignment of the C-terminal region and an indication of some of the variants used in this work are illustrated in Fig. 1b. The N-terminal regions of the proteins, corresponding approximately up to residue 390 of VP16-H, are well conserved. However, after these regions, there is a notable divergence in the sequences, indicated by a transition from yellow to green shading, where VP16-B, VP16-E, and VZV VP16 exhibit continued conservation but VP16-H does not (Fig. 1b). After this section, the C-terminal region of VP16-E is much shorter than that of VP16-H, is not noticeably acidic, and completely lacks the section around the critical phenylalanine residue 442. The exception to the general lack of homology between the C-terminal regions is found at the extreme C terminus, where a short section of approximately 15 residues is well conserved. This section is within the region of VP16-H which we previously termed H2 (46) and, for the sake of clarity, has been labeled in this work (Fig. 1b) as the H2 core conserved region (H2.CCR).

View larger version (72K): [in this window] [in a new window]   FIG. 1.   Comparison of activation by VP16-H and VP16-E. (a) Schematic representation of the similarities between VP16-H and VP16-E. The N-terminal shaded regions are well conserved, while the C-terminal regions are not, with the exception of a short section retained in both species, as discussed in the text. (b) Primary sequence alignment of the VP16 homologues extending from residue 357 in VP16-H. Yellow shading indicates the region well conserved in all species, while green shading indicates the region which appears to be selectively conserved in VP16-E, VP16-B, and VZV VP16. The H2.CCR at the extreme C termini of all species is highlighted, while the C-terminal boundaries of variants of VP16-E analyzed in this work are indicated. (c) Comparison of activation by VP16 variants. The CAT reporter construct pCAT_TAAT (8), which contains the EHV-1 IE gene promoter region (360 to +78), was transfected (200 ng) into Vero cells either alone () or together with the indicated VP16 constructs (at doses of 1, 10, and 100 ng). Cells were harvested 40 h later, and aliquots were assayed for CAT activity. (d) As for panel c but with the VP16-H constructs indicated and the target vector pAB5, which contains the HSV-1 IE110 gene promoter from 165 to +150 (30).

Differences in requirements for recruitment to the DNA binding complex. Although the C-terminal region of VP16-H is not required (10) for the assembly of the octamer binding complex (TRF-C) with Oct-1 and HCF, it was nonetheless possible that the failure of VP16-E.(1-425) to activate expression was due to a failure to be recruited into the corresponding complex. To examine this possibility, wild-type VP16-E, VP16-E.(1-425), and several additional C-terminal deletion variants were expressed in vitro and assayed for the ability to promote complex formation in gel retardation assays. Each of the proteins was expressed in vitro at approximately the same level (Fig. 2a), and equal amounts of the products were incubated with in vitro-translated HCF, the purified POU domain (0.25 ng), and the E1 octamer probe from the EHV-1 IE gene promoter (Fig. 2b). Independent binding of the POU domain to the E1 probe was observed (Fig. 2b, POU), together with a complex (asterisk) which originated from the TnT control lysate and was observed in all experiments to various degrees. The formation of TRF-C (Fig. 2b, Complex) was observed dependent upon the addition of both VP16-E and HCF (lanes 3 to 5). Either component alone failed to promote complex formation (Fig. 2b, lanes 1 and 2). Surprisingly, deletion of the C-terminal 23 residues [VP16-E.(1-425)] almost completely eliminated complex formation (Fig. 2b, lanes 7 to 9). A variant containing a further deletion to residue 393 similarly failed to promote complex formation (data not shown). Smaller deletions within and around the extreme C terminus had little effect on complex formation (Fig. 2b, lanes 10 to 25). Note that the relatively poorer complex formation observed in this experiment for the wild-type and VP16-E.(442-445) species was not a reproducible effect (see, e.g., Fig. 3; also, data not shown).

View larger version (69K): [in this window] [in a new window]   FIG. 2.   Complex formation by VP16-E and variants. (a) VP16-E variants were translated in vitro in a rabbit reticulocyte system (50 µl) in the presence of [35S]methionine, and equal amounts (1 µl) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and detected by autoradiography. Positions of molecular size markers (in kilodaltons) are indicated on the left. (b) Equal amounts of in vitro translation reactions for the species indicated (1, 2, or 4 µl) were incubated with the E1 octamer probe, the purified POU domain, and HCF, supplied from a separate in vitro translation reaction. The position of the complex dependent on all components is indicated. Independent binding of the POU domain is indicated by POU, while a nonspecific complex formed by a component of the reticulocyte extract is indicated by an asterisk. The position of unbound probe is marked as DNA. The results for each VP16 species are shown in four lanes, the first lane of which lacks HCF. Thus, complex formation by competent variants is seen to be dependent upon HCF. The control lanes (lanes 1) show that HCF is not sufficient for complex formation, which requires the appropriate VP16 species.

View larger version (36K): [in this window] [in a new window]   FIG. 3.   (a) Requirement for a determinant located between positions 425 and 445 for VP16-E complex formation. COS-1 cells were transfected (2 µg of DNA) with an expression vector for wild-type VP16-E, VP16-E.(1-445), or VP16-E.(1-425), and soluble extracts were made 40 h after transfection as described in Materials and Methods. Extracts in the amounts indicated (microliters) were incubated with the purified Oct-1 POU domain, in vitro-translated HCF, and radiolabeled EHV-1 E1 probe. Complexes were then separated on a nondenaturing TBE-6% polyacrylamide gel. The nonspecific complex formed by a component in the reticulocyte extract is indicated by a single asterisk, while a component likely representing endogenous Oct-1 in the soluble extract is indicated by a double asterisk. Free probe (DNA), POU binding, and the POU-HCF-VP16 complex are indicated by arrows. Controls in lanes 1 and 2 and the first lane of each series show that the complex was dependent upon HCF and competent VP16-E. Deletion of the 20 residues from positions 445 to 425 abolished complex formation. (b) Summary of each of the VP16-E constructs and their function in complex formation (Comp.) (this work) and IE gene transactivation (Act) (8; this work). Salient features include the lack of complex formation by VP16-E.(1-425) and the competence for complex formation of VP16-E.(442-445), despite its lack of transactivation. In this summary, activation was indicated as negative when it measured less than 10% of control values.

Cellular compartmentalization of VP16-E. To examine the compartmentalization of VP16-E in comparison to that of VP16-H, VP16-E and variants were tagged at their N termini with an epitope tag (from the SV5 matrix protein), and localization in transfected COS-1 cells was assessed by immunofluorescence. Cells were fixed in cold methanol, and VP16-E was detected with the anti-SV5 monoclonal antibody. VP16-E typically showed a significant degree of nuclear accumulation, with lower but detectable amounts in the cytoplasm, particularly in cells with a higher level of expression (Fig. 4a). Nuclear accumulation of VP16-E was more pronounced than that of VP16-H which, in a parallel analysis, was found in a diffuse, predominantly cytoplasmic pattern, as previously reported (20). Patterns for VP16-E.(1-425) and VP16-E.(1-393) exhibited no significant differences in compartmentalization relative to wild-type VP16-E (Fig. 4a), and overall levels of expression were similar (Fig. 4b). The results help support the proposal that there was no gross disruption of VP16-E by virtue of the C-terminal deletions and that the failure to promote activation in vivo was due to a failure to promote complex formation for variants VP16-E.(1-425) and VP16-E.(1-393).

View larger version (65K): [in this window] [in a new window]   FIG. 4.   Localization of VP16-E. (a) Expression vectors (1 µg) for epitope-tagged VP16-E (w/t), VP16-E.(1-425), and VP16-E.(1-393) were transfected into COS-1 cells, which were fixed and processed for immunofluorescence as described in Materials and Methods. Each VP16 species exhibited significant nuclear accumulation, with little difference between variants. (b) Levels of expression were detected in aliquots of total sodium dodecyl sulfate lysates of the transfected cells after separation by electrophoresis and Western blotting with the SV5 monoclonal antibody. Wild-type VP16-E, VP16-E.(1-425), VP16-E.(442-445), and VP16-E.(1-393) are shown in lanes 1 to 4, respectively. Positions of molecular size markers (in kilodaltons) are indicated on the left.

Activation by GAL-VP16-E. The method usually used to uncouple requirements for DNA binding (whether through protein-protein interactions, as for VP16, or via direct DNA binding) from those for transcriptional activation per se is to link the candidate activation domain to an independent DNA binding domain, for example, that of the GAL4 protein. However, since deletion of the H2.CCR had little effect on VP16-H but had a profound effect on VP16-E, it seemed likely that other determinants, possibly within the N-terminal region itself, were involved in activation by VP16-E, acting as the equivalent of the H1 region of the VP16-H C-terminal activation domain. Indeed, a region close to the N terminus of VP16-E, between residues 23 and 46 (our numbering system; previously labeled as residues 53 to 77), has been indicated to share some homology with the H1 region of VP16-H (35).

View larger version (36K): [in this window] [in a new window]

View larger version (44K): [in this window] [in a new window]   FIG. 5.   Comparison of activation by intact VP16 species versus GAL fusions. (a) Vero cells were transfected with the EHV-1 IE gene target construct pCAT_TAAT (200 ng) alone () or together with 1, 10, or 100 ng of the VP16-H and VP16-E expression vectors as indicated, and CAT activity was assayed. (b) COS-1 cells were transfected with the upstream activation sequence target construct pUAS10CAT (475 ng) alone () or together with 10, 100, or 1,000 ng of the GAL4 fusion constructs as indicated. The relative strength of VP16-E when the intact proteins were assayed is not reflected in the context of the GAL4 fusion proteins.

View larger version (60K): [in this window] [in a new window]   FIG. 6.   Lack of activation by VP16-H/E. (a) COS-1 cells were transfected with the HSV-1 IE gene reporter construct pAB5 (100 ng) alone () or together with expression vectors for VP16-E, VP16-H, and the VP16-H/E chimera [VP16-H.(1-411) plus VP16-E.(393-448)] at doses of 1, 10, and 100 ng, and CAT activity was measured. (b) Levels of expression of VP16-H and VP16-H/E from cells transfected with 100 ng of DNA were compared. Whole-cell extracts were separated on a sodium dodecyl sulfate-10% polyacrylamide gel, and the products were transferred to nitrocellulose and probed with the anti-VP16 monoclonal antibody LP1. Positions of molecular size markers (in kilodaltons) are indicated on the left.

View larger version (26K): [in this window] [in a new window]   FIG. 7.   Complex formation by VP16-H/E. (a) VP16-H and VP16-H/E were translated in vitro in reticulocyte lysates, and equal amounts of the lysates (in microliters) were incubated with purified POU in the presence or absence of in vitro-expressed HCF exactly as described in the legend to Fig. 2, with the exception that poly(dI-dC) (1 µg) was included as a nonspecific competitor. The samples were analyzed on a nondenaturing TBE-6% polyacrylamide gel. Free probe (DNA), POU binding, and the POU-HCF-VP16 complex are indicated by arrows. Low amounts of the nonspecific band from the TnT lysate are marked by an asterisk. (b) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of in vitro-translated VP16-H and VP16-H/E fusion proteins. Positions of molecular size markers (in kilodaltons) are indicated on the left. (c) Schematic summary of determinants in VP16-H and VP16-E. H1 and H2 represent redundant determinants in the C terminus of VP16-H, and H2 is a conserved determinant in VP16-E. N1 represents an N-terminal determinant in VP16-E to account for activation by native VP16-E but not by the VP16-H/E chimera, despite the ability of the latter to promote complex formation.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

VP16 is a potent activator of HSV-1 IE gene expression, the study of which has yielded significant insights into eukaryotic transcriptional regulation. From such studies, the view has emerged that VP16 comprises two independent domains, a large N-terminal domain involved in binding Oct-1, HCF, and DNA to form the DNA binding complex and a C-terminal domain involved in activation per se. To examine this view further, we have taken the approach of comparing requirements within the C-terminal regions of two related VP16 species, VP16-H and VP16-E, from HSV-1 and EHV-1, respectively, for recruitment into the DNA binding complex containing Oct-1 and HCF and for activation. The results indicate that VP16-E does not appear to conform to the previously described view of VP16-H of segregation into two independent domains. In VP16-E, a determinant within the extreme C terminus is directly involved in complex formation or somehow ensures correct presentation of the determinants in the N terminus.

The specific requirement in VP16-E for this C-terminal region in complex formation is something of a paradox. On the one hand, the extreme C termini encompass a short determinant (the H2.CCR) which is conserved, despite the overall lack of similarity in the C termini. Deletion of this determinant may therefore have been expected to have a consequence on some activity. This is the case for VP16-E, in that such deletion abolishes transactivation by affecting complex formation. On the other hand, in VP16-H, the removal of the acidic domain, including the H2.CCR determinant, has no effect on complex formation (10). Thus, for VP16-E, the view of the C terminus as being exclusively involved in some aspect of transactivation is not accurate. Notwithstanding that VP16-E retains a high degree of homology with determinants required for complex formation in the N-terminal domain of VP16-H, it appears that features within the C terminus are required for complex assembly. It is noteworthy, though, that certain residues within the H2.CCR may be specifically involved in activation, since we show that the mutant VP16-E.(442-445) retained complex-forming activity while being inactive in transactivation (8).

The overlap between determinants involved in complex formation and those required for transactivation makes analysis of structure-function relationships difficult. To examine activation distinct from complex formation, we fused the entire VP16-E protein to the DNA binding domain of GAL4 and compared activation by VP16-E and VP16-H in the native context to that in the GAL4 fusion setting. While the native proteins exhibited approximately equal activities, in the context of the GAL4 fusions, VP16-E was significantly weaker than VP16-H, exhibiting little activity. These results indicate two main points. First, the use of artificial GAL fusion proteins may not reflect the true nature of activation domains, both in qualitative and in quantitative terms. Results from this work and previous work (26) indicate that the GAL4-acidic domain fusion protein is unusually potent and is in fact considerably more potent than GAL4 fused to the entire VP16-H protein. A positive interpretation of this result could be that the extremely potent activity of the GAL4-acidic domain protein does represent the activity of the acidic domain of native VP16-H when, for example, it becomes released or exposed for full activity through its normal assembly pathway. On the other hand, it is conceivable that the mechanism involved in activation by the GAL4-acidic domain fusion protein does not truly reflect that of the native protein. An alteration in the function of VP16 in the context of a GAL4 chimera has been suggested previously (12). That work showed that VP16-H acting on its native promoter could enhance activation only when in a promoter-proximal position, while GAL-VP16-H.(411-490) could additionally activate from a distal position. A related phenomenon was observed with certain HOX proteins, where identification of the activation domains of the HOX proteins varied dramatically depending on the context of the proteins (45). Analysis of the binding of HOXD9 as a monomer to a HOX binding site located the activation domain to within residues 76 to 264. However, when the same analysis was carried out using a GAL4 DNA binding domain fused to the HOX protein, only the N-terminal 75 amino acids contained a potential activation domain; this same region, however, could be deleted in a native setting with virtually no effect (45). These and other results from similar types of analysis of other regulatory proteins indicate that caution in the interpretation of the results of studies in the context of GAL4 fusion proteins seems justifiable. It is possible that the mode of DNA binding has direct consequences on the mechanism of activation, for example, in the presentation of the activation domain. With that in mind, we proceeded to construct a VP16 chimera, but here, too, we obtained surprising results.

Our rationale was to explore similarities and differences in VP16-H and VP16-E by constructing a chimera, informed by a previous analysis of the sufficiency of the N-terminal domain of VP16-H for complex formation and our alignment of the homologues to indicate where transitions or boundaries in functional sections may occur. Furthermore, recent determination of the crystal structure of VP16-H indicates that there is little ordered structure beyond residue 345, presumably indicative of a disordered unfolded region which appears to have little influence on the folding of the N-terminal core domain (24). Thus, a reasonable expectation would be that linking the complete N-terminal region to the C-terminal region of a related protein would have little effect on folding. We selected the region of VP16-E to be linked on the basis of the alignments and the transition at residue 393 (of VP16-E) to a C-terminal section showing good conservation in the VP16-B, VP16-E, and VZV VP16 species (Fig. 1b). The chimera was expressed normally; the N terminus folded normally, as indicated by the ability of the chimeric protein to promote complex formation, yet it activated transcription extremely poorly. The salient comparison is between native VP16-E and VP16-H/E. Both retain the C-terminal region, both promote complex formation, yet VP16-E is a potent activator and VP16H/E exhibits little activity. One explanation for this result, illustrated schematically in Fig. 7c, could be that for VP16, there exist two C-terminal determinants, H1 and H2, which are redundant for transactivation, with the result that deletion of H2 has little effect. This notion would be consistent with previous results (10, 35, 42). It is reasonable to suggest that VP16-E has only one C-terminal determinant, equivalent to H2, and that deletion of this region renders the protein inactive. However, since the VP16-H/E chimera is also inactive, the explanation requires a difference between the N termini of VP16-H and VP16-E, with the latter possessing an extra determinant, N1, not present in VP16-H and required together with H2 for activation.

An N-terminal region in VP16-E (residues 23 to 46) exhibiting some similarity to H1 has been previously reported (27), and this may be the difference between the two proteins; however, we note that this region in VP16-E does exhibit conservation with the extreme N terminus of VP16-H. The determinant would also need to be qualitatively distinct from the VP16-H H1 determinant, which appears to suffice in the absence of H2, while the N terminus of VP16-E does not. An alternative explanation is that in both proteins, the N- and C-terminal determinants always function in conjunction. Thus, in VP16-H, activation could be mediated by an N1-H1 or an N1-H2 interaction, while in VP16-E, there would be only an N1-H2 determinant. This explanation, however, is further complicated by the requirement in VP16-E for H2 in complex formation itself; it is possible that in VP16-E an interaction between the N and C termini is required for complex formation, while in VP16-H the N-terminal region is sufficient. An additional chimera in which additional N-terminal regions of VP16-E are spliced into the VP16-H/E chimera may exhibit restored function and identify differences between the two proteins.

Finally, however, the basis for the clear selective conservation of the H2.CCR in each of the homologous remains to be established. In this respect, we have previously shown that VP16-H interacts with another tegument protein, VP22, that the C-terminal domain is required, and furthermore that residues in the H2.CCR are involved in this interaction (7). It is possible that the other VP16 homologues interact with the corresponding VP22 species and that the conservation within the H2.CCR is related to this activity. Moreover, it is possible that any interaction between the VP16 H2.CCR and VP22 in the virion is relevant to presentation of the activation domain early after entry, i.e., in activation, or late in infection, when activation may be suppressed. We are currently examining this proposition with cotransfection assays.

In summary, our results indicate that the simplified model of discrete domains within VP16-H for complex formation and activation may not apply to other VP16 homologues and that GAL4 fusion proteins may not reflect determinants or activities of the parental proteins. The result is that at least in certain VP16 species, the separation of determinants involved in recruitment to DNA binding complexes from those involved in activation may be more complicated than previously appreciated.